Happiness has been a central tenet of Western culture since the days of Greek philosophers (Graham 2008). Although there are many definitions of happiness, generally, it could be defined as an abstract and subjective assessment of oneself or the holistic assessment of one’s entire life (Veenhoven 2008). Happiness is often described as the main aim of life and an individual’s drive for personal fulfillment (Scoffham and Barnes 2011, Weech-Maldonado R. et al. 2017).
Nowadays, as mental health becomes one common societal issue, people’s well-being has attracted policymakers worldwide. Happiness is not a purely personal issue; it is strongly determined by the society they live in (Frey & Stutzer 2002: p. vii). On the one hand, some reported satisfaction with life across countries is positively correlated with average income. In other words, people in richer countries are likely to be happier than those in poor countries in a long term (Easterlin et al. 2010). On the other hand, there are also findings showing that income inequality, perceived unfairness, and lack of trust rather than the income level lead to more negative effects on people’s well-being (Oishi et al. 2011). Besides, Cheng et al. (2014) find more factors, including universal disposition, cultural self-construal, and national income can elucidate differences in subjective well-being through a multilevel structural equation model. Moreover, Cordero et al. (2017) claim that age, marital status, religion, and unemployment, these kinds of traditional determinants also significantly impact on efficiency of converting resources into a higher level of happiness. The arguments indicate that the factors affecting people’s well-being and the ability to transfer resources into higher level happiness are complex.
The most professional report about states’ happiness index is the World Happiness Report, distributed annually by the United Nations Sustainable Development Solutions Network based on respondent ratings of their own lives. The happiness index correlates with various life factors, such as financial generation, social back, life anticipation, flexibility, nonattendance of debasement, and liberality. The original dataset for the report provides a populated-weighted average score on a scale running from 0 to 10 for each of the variables from these six aspects, including 1) real GDP per capita, 2) social support, 3) healthy life expectancy, 4) freedom to make life choices, 5) generosity, 6) perceptions of corruption.
Global happiness report is a landmark survey of the state of global happiness and has attracted more and more attention from policymakers from many areas, including the governments, organizations, and civil society, etc. But there are two limitations of their report. Firstly, the factors listed in the report are not comprehensive. This research will explore the other factors that are not listed in the World Happiness Report and have imposed an effect on the people’s national happiness index, such as democracy level, demographic factors, Covid-19 severity (during 2020) using OLS regression models. Secondly, the annual report is based on the nearest two- or three-year data rather than panel data so that it lacks long-term analysis of the correlation between social factors with happiness scores. Thus, this research will build several fixed-effect regression models based on newly generated panel datasets to increase the validity of regression results.
In this research, I would like to explore the following questions:
Firstly, does the democracy level influence country’s happiness index?
Secondly, do the demographic factors, such as population density, population net change, and net migrants impact the national happiness score?
Thirdly, since the first identification coronavirus case in December 2019, the virus has taken the world by storm. Besides the economic impact, the pandemic period greatly impacts people’s mental health, which could not be overlooked. Thus, what are the effects of the mortality due to Covid-19 and the index of exposure to COVID-19 infections in other countries on people’s national happiness index in 2020?
Different scales of democracy will reflect the political environment inside the country. It could indicate people’s freedom to make life choices to some degree and indicate citizens’ political freedom. In a complete autocracy with the lowest democracy level, there are no multiparty elections for the chief executive or the legislature in one country (e.g., North Korea since 1945). In some electoral autocracy, which has a relatively low level of democracy (e.g., Iran since 1980, Pakistan since 2002, Turkey since 2014, etc.), even there are multiparty elections, but the elections are unfair and not free, or no multiparty elections. In some incomplete democracies with a middle level of democracy, citizens have free and fair multiparty elections in reality, but either access to justice, or transparent law enforcement, or liberal principles of respect for personal liberties, the rule of law, and judicial as well as legislative constraints on the executive are not satisfied (e.g., India since 1952 (except for 1975 and 1976), Peru during 1981-1991 and 2001-till date, Argentina since 1984, Mexico since 1995, etc.). In liberal democracy with the highest democracy level, citizens have free and fair multiparty elections and have access to justice, transparent law enforcement, and the liberal principles of respect for personal liberties, the rule of law, and judicial as well as legislative constraints on the executive (e.g., Australia since 1901, most of the European countries, United States, Canada, and Japan in the modern periods) (Coppedge et al. 2020, p. 266). Generally, when the democracy level is higher, people will feel more satisfied since they feel more engaged in political decision-making. Here comes the first hypothesis as follows:
H1: More democratic a country’s regime type is, the higher the happiness score the country will have.
Research focusing on socio-demographic factors finds that socio-demographic differences, such as gender, age, race, income, education, etc., can impact an individual’s level of happiness (Kim-Prieto et al. 2005). When comparing the population size, population density, and net population change across countries, the number of net immigrants might affect people’s satisfaction with the living environment. Specifically, in the countries with a similar amount of natural resources, economic capacity, and economic growth rate, the ones with larger population size, higher population density, rapid population net change, and higher Index of the net immigrant will have fewer natural and social resources per capita and the tendency will put more pressure on citizens so that their perception of a future life will be less likely be better compared to the people living in the other countries with smaller population size, lower population density, slow population net change, and lower Index of a net immigrant.
Thus, here come the second hypotheses:
H2a: Population size is negatively correlated with happiness index.
H2b: Population density is negatively correlated with happiness index.
H2c: Population net change is negatively correlated with happiness index.
H2d: The number of net migrants is negatively correlated with the happiness index.
H2e: Population density net change is negatively correlated with happiness index.
COVID-19, which was first discovered and reported in Wuhan, China, in December 2019, spread worldwide at a fast and terrifying pace throughout 2020. The pandemic has affected many key aspects of life around the world. The most severe impact of the pandemic is the 2 million deaths from COVID-19 in 2020. The nearly 4 percent annual increase in deaths worldwide represents a serious loss of social welfare. In terms of life, financial insecurity, anxiety, disruption in every aspect of life, stress, and challenges to mental and physical health for many people (World Happiness Report, 2021, p. 7). Countries with the highest deaths also had the greatest falls in GDP per head (p. 8). Thus, COVID-19 deaths in 2020, excess deaths in 2020 relative to the 2017-2019 average, and an Index of exposure to COVID-19 infections in other countries might harm people’s happiness index.
Here is the third hypothesis as follows:
H3: The serious the pandemic in one country, measured by COVID-19 deaths per 100,000 population in 2020 and Index of exposure to COVID-19 infections, the more decrease in the country’s happiness index at the same year.
The new datasets include Dataset 1 based on World Happiness Report 2020 and 2021 and Dataset 2 based on World Happiness Report Panel Data during 2005-2020.
This dataset is generated from the following five datasets:
Each dataset is based on the survey surveys collected from over 150 countries. The 2020 dataset features the happiness score averaged over the years 2017–2019, and the 2021 one features the happiness score averaged over 2018–2020. I have merged the two datasets and have built the difference in happiness score, difference in logged GDP per capita, difference in social support, difference in healthy life expectancy, difference in freedom to make life choices, difference in generosity, and difference in perceptions of corruption seven variables. The main variable happiness Score and the new generated difference in happiness score will be the dependent variables in my research. The Happiness Score is a national average of the responses to the main life evaluation question in the Gallup World Poll (GWP), which uses the Cantril Ladder. There are also six variables correlated to happiness score, each of which is measured reveals a populated-weighted average score on a scale running from 0 to 10, including 1) real GDP per capita, 2) social support, 3) healthy life expectancy, 4) freedom to make life choices, 5) generosity, 6) perceptions of corruption. These six metrics are used to explain the extent to which each factor contributes to increased life satisfaction when compared to the hypothetical nation of Dystopis, which represents the lowest national averages for each key variable.
The Polity5 dataset covers all major, independent states in the global system from 1800-2018 (Marshall and Gurr 2020). I will utilize the Polity2 variable to measure one state’s democracy. The Polity conceptual scheme is unique because it examines concomitant qualities of democratic and autocratic authority in governing institutions rather than discreet and mutually exclusive forms of governance. This perspective envisions a spectrum of governing authority that spans from fully institutionalized autocracies through mixed, or incoherent, authority regimes (termed “anocracies”) to fully institutionalized democracies. The “Polity Score” captures this regime authority spectrum on a 21-point scale ranging from -10 (hereditary monarchy) to +10 (consolidated democracy). The Polity scores can also be converted into regime categories in a suggested three-part categorization of “autocracies” (-10 to -6), “anocracies” (-5 to +5 and three special values: -66, -77 and -88), and “democracies” (+6 to +10).
This dataset was created by Tanu N Prabhu using web scraping based on the data shown on the website named worldometer, which is a real-time monitoring and reporting website about the world and national population from many demographic aspects. The dataset is posed on the Kaggle website, containing the variables Population (2020), Density (P/Km2), Land Area (Km²), Population Net Change, and Migrants (net) I want to use to explore the impact of demographic factors on Happiness Score.
This dataset is downloaded from the World Happiness Report 2021 website and has been included as the appendices data to explore the effect of COVID-19 on people’s well-being. There are several variables to measure the COVID-19. I utilize the COVID-19 deaths per 100,000 population in 2020 and Index of exposure to COVID-19 infections in other countries to explore how COVID-19 affects people’s well-being in 2020. This dataset also includes the population of 2019 and 2020, used to generate the population density in 2019 combined with the variables in the population dataset.
This dataset is generated from the following four datasets:
Like each World Happiness Report 2021 dataset, it is based on the survey collected from people in over 150 countries. The period of the panel data is from 2005 to 2020. Besides the main variable happiness Score, there are also six variables correlated to happiness score – real GDP per capita, social support, healthy life expectancy, freedom to make life choices, generosity, and perceptions of corruption.
The same dataset as the one used in Dataset 1 with the same name; see data descriptions above.
This dataset is downloaded from the World Bank. Its period is from 2004 to 2020, and it includes the data of 236 countries and dependencies. The population variable measures the total population size. In other words, it counts all residents regardless of legal status or citizenship.
This dataset is also downloaded from the World Bank. The period of the dataset is from 2004 to 2020. And it includes the population density data of 236 country and dependencies. As for the population density variable, its unit is people per sq. km of land area.
In this research, I build two datasets – Dataset 1 based on World Happiness Report 2020 and 2021, and Dataset 2 based on World Happiness Report Panel Data during 2005-2020. I use Pandas and NumPy packages in python to clean, merge, sort the datasets. I build several new variables to measure the differences in happiness score, GDP per capita, social support, healthy life expectancy, freedom to make life choices, generosity, perceptions of corruption, democracy, population, and population density between 2019 and 2020 in Dataset 1. I also use diff() function to generate the first difference for all variables in Dataset 2.
Based on Dataset 1, I build five OLS linear regression models to test the factors affecting country’s happiness score in 2020 and differences in happiness score between 2018-2020 and 2017-2019 periods. Based on Dataset 2, I first build two fixed-effect regression models to test the factors affecting people’s national happiness scores by fixed year effect. Then I build two first-difference regression models to test the correlation between the first difference in every explanatory variable and the first difference in happiness score.
To visualize the correlation between the explanatory variables with the happiness score, and the first difference in the explanatory variables with the first difference in happiness score, I use R, shiny, and D3 to plot the histogram of variables used in the two datasets, the scatter plots with regression lines, etc.
Table 1 shows the statistical results of five linear regression models. In models 1-3, I use the Happiness Score in the 2021 report as the dependent variable. Model 1 is the basic model which contains the six factors included in the World Happiness Report 2021, including 1) GDP per capita, 2) social support, 3) healthy life expectancy, 4) freedom to make life choices, 5) generosity, 6) perceptions of corruption. Model 2 contains the democracy variable and five demographic variables, including population size (logged), population density (logged), population change (net), migrants (net), population density net change besides the six basic variables included in Model 1. Model 3 includes two variables measuring the severity of COVID-19 – COVID-19 deaths per 100,000 population in 2020 and the Index of exposure to COVID-19 infections in other countries besides the variables in Model 2.
In models 4-5, I use Change in Happiness Scores in the 2021 report compared to those in the 2020 report as the dependent variable. In other words, I use the difference in average happiness score during 2018-2020 and average happiness score during 2017-2019 as DV. As for the explanatory variables, I have included change in GDP per capita (logged), change in social support, change in healthy life expectancy, change in freedom to make life choices, change in generosity, change in perceptions of corruption, and change in population density between 2019 and 2020 in Model 4. I have added three demographic variables – population change (net), migrants (net), and population density net change and two COVID-19 severity measurement variables – COVID-19 deaths per 100,000 population in 2020 and Index of exposure to COVID-19 infections in other countries in Model 5 besides the variables included in Model 4.
Model 1 provides support to the correlations between the five factors included in the World Happiness Report 2021. Specifically, there are positive correlations between the variables – GDP per capita (logged), social support, healthy life expectancy, freedom to make life choices, and the dependent variable happiness score; And there is a negative correlation between perceptions of corruption and happiness score. However, there is no statistical significance for the coefficient of generosity. Model 2 shows a similar relationship between happiness score and the basic six factors, as shown in Model 1. Besides, the coefficient of democracy is positively significant at a 99% confidence level, indicating that higher democracy has a positive effect on happiness score and supports H1. However, the coefficients of the demographic variables in Model 2 are all non-statistically significant. According to Model 3, similarly, democracy is positively correlated with happiness score at the 90% confidence level, supporting H1. Besides, there is no statistical significance of demographic variables. Also, two Covid-19 severity variables are not statistically significant. However, generosity is positively correlated with the happiness score in Model 3.
Thus, we could say that the improvement in the economy, social support, healthy life expectancy, freedom to make life choices, and generosity could positively affect people’s well-being. At the same time, the increase of perceptions of corruption could negatively impact people’s well-being. Democracy level is positively correlated to happiness score, while the demographic factors do not have an obvious relationship with happiness score based on the 2020 dataset.
As for the regression results of Model 4, there is positive statistical significance for change in social support, change in healthy life expectancy, and change in freedom to make life choices, showing that the improvement in social support, healthy life expectancy, and the freedom in making life choices could be good at increasing people’s happiness index in 2020. In Model 5, change in social support and the freedom to make life choices are still statistically significant. Still, there is no evidence to show the effect of both demographic and Covid-19 severity variables. Combining the results of Model 4 and Model 5, we could summarize that change in social support plays the most significant role in improving people’s well-being in 2020. Besides, improving people’s healthy life expectancy and providing more freedom are also good choices to improve people’s well-being.
Overall, we could conclude that GDP per capita (logged), social support, healthy life expectancy, freedom to make life choices, generosity, and democracy level positively correlate to one country’s average happiness index. Improvement in social support, healthy life expectancy, and freedom to make life choices could positively affect people’s well-being. All demographic variables and COVID-19 severity variables have no significant effect on either the happiness index or the change in the happiness index. However, adding demographic variables increases R2 and adjusted R2 to some degree, indicating some correlation between demographic variables and the country’s happiness index. I will explore further using Dataset 2 – the panel data.
Table 2 shows the statistical results of two fixed effects and two first difference regression models using the panel data. In models 1 and 2, I use Happiness Score as the dependent variable. Model 1 includes the six basic factors. Model 2 adds democracy variable and two demographic variables – population size (logged) and population density (logged). In models 3 and 4, I use the Difference in Happiness Score as the dependent variable. In Model 3, the explanatory variables are the first difference of the six basic factors, including difference in GDP per capita (logged), difference in social support, difference in healthy life expectancy, difference in change in freedom to make life choices, difference in generosity, difference in perceptions of corruption, and difference in population density. Model 4 add difference in democracy, difference in population, and difference in population density based on Model 3.
Model 1 provides support to the correlations between the six factors included in the World Happiness Reports. Specifically, the first five explanatory variables – GDP per capita (logged), social support, healthy life expectancy, freedom to make life choices, and generosity are positively related to the dependent variable happiness score; And perceptions of corruption are negatively related to happiness score. Model 2 shows a similar relationship between happiness score and the basic six factors, as shown in Model 1. Besides, the coefficient of democracy is positively significant at a 99% confidence level, showing that the higher level of democracy, the higher the happiness score holding the other factors constant, providing support to H1. Moreover, population size is positively correlated to happiness score at the 99% confidence level, against H2a. And population density is negatively related to happiness score at 99% confidence level, supporting H2b. In Model 3, the results show a similar tendency as shown in the first two fixed-effect models. The results in Model 4 provide a further meaningful reference for H2. The coefficients of difference in population and difference in population density are negatively statistically significant, providing support to H2c and H2e.
In a nutshell, we could conclude that GDP per capita (logged), social support, healthy life expectancy, freedom to make life choices, generosity, and democracy level are positively correlated to one country’s average happiness index. And the relationship holds when taking the first differences of the six variables and the happiness score. Democracy level is positively correlated to happiness score, providing support to H1. As for demographic variables, population density, change in population size, and population density is negatively correlated to the happiness score or first difference of happiness score, support H2b, H2c, and H2e. However, population size is positively related to happiness score, which is against H2a. There are no other statistically significant results to test H2d and H3.